Post on 11-Feb-2022
Determinants of Levered PortfolioPerformance
Robert M. Anderson∗ Stephen W. Bianchi†
Lisa R. Goldberg‡
University of California at Berkeley
April 21, 2014§
Forthcoming in Financial Analysts Journal
Abstract
The cumulative return to a levered strategy is determined by five elements thatfit together in a simple, useful formula. A previously undocumented element is thecovariance between leverage and excess return to the fully invested source portfoliounderlying the strategy. In an empirical study of volatility-targeting strategies overthe 84-year period 1929–2013, this covariance accounted for a reduction in returnthat substantially diminished the Sharpe ratio in all cases.
Key terms: Leverage; Sharpe ratio; source portfolio; trading cost; financing cost;unintended market timing; magnified source return; excess borrowing return; riskparity; pension fund; hedge fund; fixed leverage; dynamic leverage; volatility target
∗Department of Economics, 530 Evans Hall #3880, University of California, Berkeley, CA 94720-3880,USA, email: anderson@econ.berkeley.edu.†Department of Economics, 530 Evans Hall #3880, University of California, Berkeley, CA 94720-3880,
USA, email: swbianchi@berkeley.edu.‡Department of Statistics, University of California, Berkeley, CA 94720-3880, USA, email:
lrg@berkeley.edu.§This research was supported by the Center for Risk Management Research at the University of Cali-
fornia, Berkeley. We are grateful to Patrice Boucher, Dan diBartolomeo, Claude Erb, Ralph Goldsticker,Nick Gunther, Rodney Sullivan, Betsy Treynor, Jack Treynor, Barton Waring and Sorina Zahan and twoanonymous referees for insightful comments on the material discussed in this article. Earlier versions ofthis paper circulated under the title “The Decision to Lever.”
1
1 A Simple Two-Period Example
In this paper, we show that there are five elements that determine the cumulative return to
a levered strategy, and they fit together in a simple, useful formula. Looking backward, our
formula can be used to attribute performance. Looking forward, an investor can populate
our formula with his or her forecast of the five determinants in order to generate a forecast
for return to the levered strategy.
A levered strategy begins with a fully-invested source portfolio, such as unlevered
risk parity, unlevered minimum variance, or unlevered bonds. The source portfolio is then
levered according to a leverage rule. The most common leverage rules target volatility: they
estimate the current volatility of the source portfolio in some way, and then choose leverage
so that the estimated volatility of the levered strategy matches the target. Since the
source portfolio typically exhibits variable volatility, volatility targeting requires dynamic
leverage, even if the volatility target is fixed.
Much of our intuition about levered strategies comes from single-period models. In
a single-period model, the return of the levered strategy is determined by the return of
the source portfolio, the leverage, and the financing cost associated with the leverage. By
definition, leverage is constant; there is no trade and hence no trading costs; and there is
no compounding to take into account.
Now consider a simple two-period model. Assume that the source portfolio earns a
10% arithmetic return in period 1 and a -10% arithmetic return in period 2. We invest
$100.00 which is worth $110.00 at the end of period 1 and $99.00 at the end of period 2,
as shown in Table 1. The average of the arithmetic return over the two periods is zero,
but the cumulative return of the source portfolio over the two periods is:
99− 100
100= −0.01 = −1.00%
The average arithmetic return of the source portfolio return must be corrected for com-
pounding. As noted by Booth and Fama (1992) and discussed in Appendix D, the cor-
rection subtracts half the variance of arithmetic return each period; we call this corection
the variance drag. Note that the variance of the arithmetic returns is:
(0.1− 0)2 + (−0.1− 0)2
2= .01 = 1.00%
If we subtract half the variance from the arithmetic return each period, we get a total
return of:
0.1− .01
2+ (−0.1)− .01
2= −0.01 = −1.00%
which matches the actual cumulative return over the two periods.
2
Table 1: Strategies in the Two-Period Example
Time Source Return Assets Debt Strategy Value
rS A ($) D ($) A − D ($)
Source Portfolio:
Beginning of 1 100.00 0.00 100.00End of 1 10% 110.00 0.00 110.00End of 2 -10% 99.00 0.00 99.00
Fixed Leverage Strategy:
Beginning of 1 250.00 150.00 100.00End of 1 10% 275.00 150.00 125.00End of 1′ 312.50 187.50 125.00End of 2 -10% 281.25 187.50 93.75
Dynamically Levered Strategy:
Beginning of 1 200.00 100.00 100.00End of 1 10% 220.00 100.00 120.00End of 1′ 360.00 240.00 120.00End of 2 -10% 324.00 240.00 84.00
Table 1: Calculation of return of the source portfolio and levered strategies in the two-
period model. The rows with time End of 1 represent the levered strategy prior to rebalanc-
ing, while the rows with time End of 1′
represents the levered strategy after rebalancing.
Now, consider a levered strategy. Suppose for simplicity that leverage can be financed
at the risk-free rate, which happens to be zero. We initially invest $100.00. Suppose that
we target a fixed volatility of 12% per period, and our estimate of the source portfolio
volatility is 6% at the beginning of the first period and 4% at the beginning of the second
period. Thus, we choose leverage λ1 = 126
= 2 in the first period and λ2 = 124
= 3 in the
second period.
If we try to squeeze this into a one-period analysis, we might be tempted to assume
the return will be similar to that of a strategy with fixed leverage λ = λ1+λ22
= 2.5, since
this is the average leverage over the two periods. Thus, we might expect to earn about
2.5 times the average arithmetic return of the source portfolio, or 2.5× 0.00% = 0.00%. If
we are a little more sophisticated and take compounding into account, we might expect
to earn 2.5 × −1.00% = −2.50%. However, both these answers are wrong, even for the
case of fixed leverage, and they are particularly wrong for the case of dynamic leverage.
Consider a fixed leverage strategy that uses leverage 2.5 in both periods. As noted
in Table 1, we hold assets of $250.00 in the source portfolio (financed by our $100.00 and
$150.00 in debt) at the beginning of the first period. At the end of the first period, our
assets are worth $275.00, and our debt is $150.00, so the value of the levered strategy is
3
$125.00. Even though we want to maintain fixed leverage, we need to rebalance. We hold
assets of $125.00× 2.5 = $312.50 in the source portfolio. We must borrow, increasing our
debt to $312.50-125.00=$187.50 to finance the position. At the end of the second period,
our assets in the source portfolio are worth $312.50 × 0.9 = $281.25, so the value of the
levered strategy is $281.25− 187.50 = $93.75; our cumulative return over the two periods
is 93.75−100100
= −0.625, a loss of 6.25%. The variance of levered strategy return is:
(.25− 0)2 + (−0.25− 0)2
2= .0625 = 6.25%
The correction for compounding (variance drag) is 6.25%2
per period; over two periods,
this gives -6.25%, exactly matching the realized return. Note that the variance drag is
quadratic in leverage, so constant leverage of 2.5 increases the variance drag by a factor
of 6.25.
With our more sophisticated understanding of the quadratic nature of the variance
drag, we might expect the dynamically levered strategy to have a cumulative return of
about -6.25%. However, that answer is also wrong. As shown in Table 1, the dynamically
levered strategy holds assets of $200.00 in the source portfolio at the beginning of the
first period, financed by our initial $100.00 and debt of $100.00. At the end of the first
period, the assets are worth a total of $220.00, the debt is still $100.00, so the value of
the levered strategy is $120.00. We rebalance to achieve the prescribed leverage. Since
the levered strategy calls for leverage λ2 = 3, we borrow an additional $140.00 for total
debt of $240.00, and hold assets of $360.00 in the source portfolio. We incur trading costs,
which for simplicity we assume to be zero. At the end of the second period, our shares
of the source portfolio are worth $360.00 × 0.9 = $324.00; since we owe $240.00, so our
equity is $84.00. The cumulative return to the levered strategy over the two periods is:
84− 100
100= −.16 = −16.00%
Rather than breaking even, or losing 2.5%, or losing 6.25% as we expected from our
single-period intuition, we have lost 16%.
We went wrong because we ignored the covariance between leverage λ and source
portfolio return rS:
cov(λ, rS) =(λ1 − λ)× (rS1 − 0) + (λ2 − λ)× (rS2 − 0)
2
=(2− 2.5)(.1− 0) + (3− 2.5)(−0.1− 0)
2
=−0.05− 0.05
2
= −0.05 = −5.00%
The covariance term reduces return by 5.00% each period, producing a return of
−10.00%. The arithmetic return of the dynamically levered strategy is 0.2 in the first
4
period and -0.3 in the second periods, so the variance of the return of the levered strategy
is:
(.2− (−0.05))2 + (−0.3− (−0.5))2
2=
(0.25)2 + (−0.25)2
2= 0.0625 = 6.25%
The variance drag for the dynamically levered strategy is the same as for the fixed lever-
age strategy:1 half the variance, or 3.125%, each period. Combining the covariance and
variance, we get a return of −8.125%, per period, suggesting a cumulative loss of 16.25%,
close to the cumulative actual return of −16.00% over the combination of the two periods.
As this example indicates, the covariance term can make a big difference over a few
periods. One might be tempted to think the covariance term would wash out over time.
If that were true, the covariance term might not be particularly important. Strikingly, we
found that the covariance term made a substantial difference over a very long horizon. Our
empirical examples include as source portfolios risk parity (with asset classes consisting
of US stocks and US Treasury bonds) and Treasury bonds alone, with two different types
of volatility targeting and two different volatility targets. In all of our examples, the
covariance term turned out to be negative, subtracting between 0.64% and 4.23% from
annualized return over an 84-year horizon. Consequently, the Sharpe ratios of volatility-
targeting strategies were diminished relative to their source portfolios and fixed leverage
benchmarks.
2 Synopsis of Theoretical Contributions and Empir-
ical Findings
Hedge funds, real estate investment trusts, and many other investment vehicles routinely
use leverage. Even among the most conservative and highly regulated investors such as US
public pension funds, the use of levered investment strategies is widespread and growing.2
In the period since the financial crisis, strategies such as risk parity that explicitly lever
holdings of publicly traded securities have emerged as candidates for these investment
portfolios.3
In the single-period Capital Asset Pricing Model (CAPM), the market portfolio is
the unique portfolio of risky assets that maximizes the Sharpe ratio. Leverage serves only
as a means to travel along the efficient frontier. Both excess return and volatility scale
linearly with leverage, and a rational investor will lever or de-lever the market portfolio
in accordance with his or her risk tolerance.
Empirically, certain low-volatility portfolios have exhibited higher Sharpe ratios than
did the market portfolio,4 which suggests that levering a low-volatility source portfolio
1This is true in this specific example, but is not true in general.2See, for example, Kozlowski (2013).3Sullivan (2010) discusses the risks that a pension fund incurs by employing a levered strategy.4See, for example, Anderson et al. (2012).
5
could deliver an attractive risk-return tradeoff. However, market frictions such as the
difference between borrowing and lending rates, and the correlations that arise in multi-
period models make the relationship between the realized return of a levered strategy
and the Sharpe ratio of its source portfolio both nuanced and complex. Levered strategies
tend to have substantially higher transaction costs5 than do traditional strategies.6
We develop an exact performance attribution for levered strategies that takes market
frictions into account. Specifically, we show that there are five important elements to
cumulative return. The first element is the return to the fully invested portfolio to be
levered, which we call the source portfolio. The second element is the expected return to
the source in excess of the borrowing rate, amplified by leverage minus one. We call the
sum of these terms the magnified source return, and it represents the performance of a
levered strategy in an idealized world.
In the real world, the magnified source return is enhanced or diminished by the
covariance between leverage and excess borrowing return, which is the third element of
cumulative return of a levered strategy. Empirically, the covariance term turned out to be
unstable at medium horizons of three to five years. Looking back, this made certain levered
strategies appear particularly appealing at some times and particularly disappointing at
other times. Viewed prospectively, it added considerable noise to medium horizon returns.
The fourth and fifth elements, the cost of trading and the variance drag, are familiar to
many investors. We penalized trading according to a linear model and we estimated the
variance drag, which is effectively the difference between arithmetic and geometric return,
using a formula that is adapted from Booth and Fama (1992).
Section 3 provides the foundation for our performance attribution, which is derived
in Section 3.1. In Section 3.2, we illustrate the performance attribution in the context
of a particular risk parity strategy, UVT60/40, which targeted a fixed volatility equal to
the realized volatility of a 60/40 fixed mix over our 84-year sample period, 1929–2012.
As shown in Table 2, all five terms in the performance attribution contributed materially
to the cumulative return of UVT60/40. For example, the covariance term subtracted an
average of 1.84% per year from the expected arithmetic return of the magnified source
portfolio.
Section 4 discusses the assumptions we made about historical borrowing and trading
costs and their impact on performance comparisons.
Our performance attribution facilitates a comparison between a levered strategy and
a variety of benchmarks, which are explored in Section 5. The benchmarks fall into two
classes. The first consists of fully invested portfolios, while the second consists of portfolios
5Investment returns are often reported gross of fees and transaction costs. That practice may be
reasonable in comparing strategies with roughly equal fees or transaction costs, but it is inappropriate
when comparing strategies with materially different fees or transaction costs.6By traditional strategies, we mean the strategies that have typically been employed over the last 50
years by pension funds and endowments. These strategies invest, without leverage, in a relatively fixed
allocation among asset classes.
6
that use fixed leverage. For example, we compare UVT60/40 to its source portfolio and its
volatility target, a 60/40 fixed mix. The comparison of a levered strategy to fully invested
benchmarks is important since there would be no rational reason to invest in a levered
strategy if it underperformed these benchmarks. However the comparison is clouded by
the fact that backtests of levered strategies rely on assumptions about historical financing
costs, while backtests of unlevered strategies do not.7
By contrast, comparisons among backtests of different types of levered strategies
are on firmer ground: even if there are errors in the assumptions about financing costs,
they affect all the strategies under consideration in similar ways. We introduce two fixed-
leverage strategies in Section 5.2. The first, FLT60/40,λ had constant leverage equal to
the average leverage of UVT60/40; the second, FLT60/40,σ, had constant leverage and had
volatility equal to the volatility of UVT60/40. In our backtests, the fixed leverage strategies
outperformed UVT60/40 as well as a conditional volatility-targeting risk parity strategy,
CVT60/40, which is also introduced in Section 5.2. The volatility-targeting strategies had
lower Sharpe ratios than the corresponding fixed-leverage strategies, which had lower
Sharpe ratios than the underlying source portfolios. Section 5.3 discusses how the levered
strategies UVT, CVT, and FLT responded to changes in market conditions; in particular,
with λ > 1, these turned out to be momentum strategies.
CVT60/40 matched the contemporaneous volatility of the fixed-mix 60/40, rather than
its unconditional volatility over a long horizon. An advantage of CVT60/40 over the other
strategies is that is investable: perfect foresight is not required to rebalance the strategy
each month. On the other hand, UVT and FLT strategies can be set by choosing a fixed
volatility or leverage that is in the ball park of the expected future volatility of the target.
This raises the question of sensitivity to parameters: if we set a UVT target volatility with
an intent to match the volatility of a given strategy, such as the value-weighted market or
60/40, how close will the performance of the strategy we implement be to the performance
of the strategy we intended to implement? We do not seriously address this question here,
but a hint about its complexity and depth is in Section 5.4, which looks at the impact
of the target volatility on strategy performance. These four risk parity strategies lever a
common source portfolio, so it is straightforward to compare the return attributions of the
strategies. The details are in Table 6, which shows, for example, that the covariance drag
in UVT60/40 was substantially larger than in CVT60/40, and the difference in the covariance
drags of UVTMKT and CVTMKT was even more pronounced. The high magnitude of the
covariance drag and its sensitivity to the volatility target in UVT, came from both a high
volatility of leverage and high sensitivity of the volatility of leverage to the UVT volatility
target, compared to CVT.
In Section 5.5, we look beyond risk parity by considering a US government bond index
levered to the volatility of US equities. The results are qualitatively similar although they
7To the extent that levered strategies exhibited higher turnover than fully invested strategies, their
returns may have been more sensitive to assumptions about historical trading costs. In our empirical
results, financing costs had a significantly greater impact than did trading costs.
7
are more dramatic since the volatility of the source portfolio is lower in this example
than in the others, while the target volatility is higher. The results are in Table 7. For
example, the covariance term in UVTBSTOCKS subtracted 4.23% per year from strategy
performance.
Section 6 revisits the covariance term from the viewpoint of volatility-targeting. It
demonstrates that the covariance term is still present from the volatility-matching per-
spective, and demonstrates that fixed-volatility targeting is a form of unintended market-
timing, whereas fixed leverage is not.
In all of the volatility-targeting strategies we considered, the covariance term in the
UVT strategies was negative over our 84-year data period. We note that the UVT co-
variance term was positive over many three-to-five year periods, and some periods lasting
two to four decades.
Section 7 summarizes our main conclusions.
We also include a number of appendices that support our main narrative. Appendix A
provides a detailed overview of the literature on low-risk investing and leverage. Ap-
pendix B describes the data in enough detail to allow researchers to replicate our results.
Appendix C describes our linear trading model. Appendix D derives our approximation
of geometric return from arithmetic return. As illustrated in our empirical examples, this
approximation has a high degree of accuracy in practical situations. Appendix E presents
a table with the formulas and corresponding words for the elements of our performance
attribution.
3 The Impact of Leverage on the Return to an In-
vestment Strategy
Leverage magnifies return, but that is only one facet of the impact that leverage has
on an investment strategy. Leverage requires financing and exacerbates turnover, thereby
incurring transaction costs. It amplifies the variance drag on cumulative return due to
compounding. When leverage is dynamic, it can add substantial noise to strategy re-
turn. We provide an exact attribution of the cumulative return to a levered strategy that
quantifies these effects.
A levered strategy is built from a fully invested source portfolio of risky assets, pre-
sumably chosen for its desirable risk-adjusted returns, and a leverage rule.8
An investor has a certain amount of capital, L. The investor chooses a leverage ratio
λ, borrows (λ− 1)L, and invests λL in the source portfolio.9
In what follows, we assume λ > 1.
8The source portfolio can be long-short in the risky assets. It must, however, have a non-zero value,
so that returns can be calculated. Since we want to model leverage explicitly, we do not allow the source
portfolio to contain a long or short position in a riskless asset, such as T-bills, the money market account,
or commercial paper.9Leverage may be achieved through explicit borrowing. It may also be achieved through the use of
8
3.1 Attribution of Arithmetic and Geometric Return
The relationship between the single-period return to a levered portfolio, rL, and to its
source portfolio, rS, is given by:
rL = λrS − (λ− 1)rb, (1)
where the borrowing rate, rb, is greater than or equal to the risk-free rate rf . Note that
the excess return is given by:
rL − rf = λrS − (λ− 1)rb − rf
= λ(rS − rf
)− (λ− 1)
(rb − rf
)(2)
Excess return and volatility scale linearly in λ for λ ≥ 0 if and only if rb = rf ; in that
case, the situation is essentially the same as the single-period CAPM, except that the
source portfolio need not be the market portfolio.
When rb > rf , volatility still scales linearly in λ ≥ 0 but Formula (2) indicates that
excess return scales sublinearly; as a consequence, the Sharpe ratio is a declining function
of λ. Note that the excess borrowing return of the levered strategy is:
rL − rb = λ(rS − rb
)(3)
It is the excess borrowing return and volatility that scale linearly in leverage, for λ ≥ 1.
The bar for leverage to have a positive impact on return has gotten higher: the excess
borrowing return, rS − rb, must be positive.
The expected return to a levered strategy is estimated by rewriting Formula (3) as:
rL = rS + (λ− 1)(rS − rb
)(4)
and taking the expectation over multiple periods:
E[rL]
= E[rS]
+ E[(λ− 1)
(rS − rb
)]= E
[rS]
+ E [λ− 1]E[rS − rb
]+ cov(λ, rS − rb) (5)
We use the term magnified source return to denote the sum of the first two terms on
the right side of Formula (5). That formula shows that the expected return to a levered
strategy is equal to the magnified source return plus a covariance correction. We find
empirically that, even when the correlation between leverage and excess borrowing return
is quite small, the covariance correction can be substantial in relation to the magnified
source return.
derivative contracts, such as futures. In these derivative contracts, the borrowing cost is implicit rather
than explicit, but it is real and is typically at a rate higher than the T-Bill rate. For example, Naranjo
(2009) finds that the implicit borrowing cost using futures is approximately the applicable LIBOR rate,
applied to the notional value of the futures contract.
9
We can interpret the expectation and covariance in Formula (5) in two ways. Prospec-
tively, they represent the expectation and covariance under the true probability distribu-
tion. Retrospectively, they represent the realized mean and realized covariance of the
returns.10
Also important over multiple periods is the cost of trading, which imposes a drag
rTC on any strategy: To take account of this effect, we extend Formula (5):
E[rL]
= E[rS]
+ E [λ− 1]E[rS − rb
]+ cov(λ, rS − rb)− E
[rTC
]= E
[rS]
+ E [λ− 1]E[rS − rb
]+ cov(λ, rS − rb)−
(E[rTCS
]+ E
[rTCL
])(6)
where rTC is expressed as a sum of trading costs due to turnover in the source portfolio
and trading costs due to leverage-induced turnover:
rTC = rTCS + rTCL.
Estimates of rTC and its components rely on assumptions about the relationship
between turnover and trading cost. We assumed that cost depended linearly on the dollar
value that turned over, and we used Formulas (15) and (16) to estimate rTC in our
empirical studies. More information is in Appendix C.
Formula (6) is based on arithmetic expected return, which does not correctly account
for compounding. The correction for compounding imposes a variance drag on cumulative
return that affects strategies differentially; for any given source portfolio, the variance drag
is quadratic in leverage. If a levered strategy has high volatility, the variance drag may
be substantial.
If we have monthly returns for months t = 0, 1 . . . , T − 1 the realized geometric
average of the monthly returns is:
G[r] =
(T−1∏t=0
(1 + rt)
)1/T
− 1 (7)
where rt is the arithmetic return in month t. Given two strategies, the one with the higher
realized geometric average will have higher realized cumulative return. In Appendix D,
we show that the following holds to a high degree of approximation:11
G [r] ∼ (1 + E [r]) e−var(r)
2 − 1 (8)
Note that the correction depends only on the realized variance of return.12 Booth and
Fama (1992) provide a correction for compounding based on continuously compounded
10Note that we take the realized covariance, obtained by dividing by the number of dates, rather than
the realized sample covariance, which would be obtained by dividing by one less than the number of
dates. We use the realized covariance because it makes Formula (5) true.11The magnitude of the error is estimated following Formula (20). Note thatG and E denote realizations
of the geometric and average arithmetic return, respectively. The term var(r) denotes the realized variance
of r, rather than the realized sample variance.12In an earlier version of this paper, we indicated, incorrectly, that both the level and the variability
of volatility determine the magnitude of the variance drag.
10
return; our correction for the geometric average of monthly returns in Formula (8) is
slightly simpler.
Thus, in comparing the realized returns of strategies, the magnified source return of
the levered strategy must be adjusted for three factors that arise only in the multi-period
setting: the covariance correction, the variance drag, and trading costs.13
3.2 Empirical Example: Performance Attribution of a Levered
Risk Parity Strategy
We demonstrate the utility of the performance attribution detailed above in the context
of UVT60/40, a risk parity strategy that was rebalanced monthly and levered to an uncon-
ditional volatility target equal to the realized volatility, 11.59%, of the 60/40 fixed-mix
between January 1929 and December 2012.14 The source portfolio was unlevered risk par-
ity based on two asset classes, US Equity and US Treasury Bonds. Foresight was required
in order to set this target: the volatility of the 60/40 strategy was not known until the
end of the period.15
Figure 1 shows the magnified source return and the realized cumulative return to
UVT60/40, as well as the realized cumulative return to its source portfolio (fully invested
risk parity) and target (60/40 fixed mix). All computations assumed that leverage is
financed at the 3-month Eurodollar deposit rate. The realized cumulative returns were
based on the additional assumption that trading is penalized according to the linear
model described in Appendix C, and took into account the covariance correction and
variance drag on cumulative return. The magnified source return of UVT60/40 easily beat
the cumulative return of both the source and the target; however, the realized cumulative
return of UVT60/40 was well below the realized cumulative return of the 60/40 target
portfolio (with essentially equal volatility (11.58%)) and only slightly better than unlevered
risk parity source portfolio, which had much lower volatility (4.20%).16
13Note that the source and target portfolios may incur their own trading costs, as well as benefit from
volatility pumping. The performance attribution of Formula (6) uses the source return and magnified
source return, gross of trading costs. When we report historical arithmetic returns to the source and
target portfolio, we report these net of trading costs, and inclusive of any benefit from volatility pumping.
When we report cumulative returns to the source and target portfolios, we report these net of the variance
drag.14The leverage was chosen so that the volatility, gross of trading costs, was exactly 11.59%. When
trading costs were taken into account, the realized volatility was slightly lower: 11.54%. UVT60/40 was
constructed in effectively the same way as the levered risk parity strategy in Asness et al. (2012), with
one main difference. They levered risk parity to match the volatility of the market, which had higher
volatility than 60/40. In Section 5.4, we consider risk parity levered to the volatility of the market.15The sensitivity of strategy performance to the volatility target is discussed in Section 5.4.16The volatilities are reported in Table 3.
11
0.1
1
10
100
1000
10000Source: Risk Parity, Target: 60/40 (1929−2012)
1937
0930
1946
0628
1955
0331
1963
1231
1972
0929
1981
0630
1990
0330
1998
1231
2007
0928
Src
Targ
Mag Src
UVT60/40
Figure 1: Magnified source return (in magenta) and realized cumulative return (in light
green) for UVT60/40 (risk parity unconditionally levered to a target volatility of 11.59%)
over the period 1929–2012. For comparison, we also plot the realized cumulative return of
the volatility target (60/40 fixed mix, in blue) and the source (fully invested risk parity,
in lavender). Magnified source return is an idealized return that cannot be achieved in
practice; the curve depicts what we would earn if we achieved a geometric return equal
to the arithmetic magnified source return.
The return decomposition Formulas (6) and (8) provide a framework for analyzing
the performance of UVT60/40. Table 2 provides the required information. Consider first
the magnified source return. The source portfolio had an annualized arithmetic return of
5.75% gross of trading costs.17 Leverage added an extra 3.97% to annualized return from
the magnification term, the average excess borrowing return to the source portfolio mul-
tiplied by average leverage minus one. The annualized magnified source return was thus
9.72%. However, the covariance between leverage and excess borrowing return reduced
the annualized return by 1.84%, trading costs by 96 basis points, and variance drag by a
17Trading costs subtracted only 7 basis points per year from the source return.
12
further 48 basis points. Together, these three effects ate up 3.28% of the 3.97%, or 82.6%,
of the contribution of leverage to the magnified source return.
Table 2: Performance AttributionSample Period: 1929-2012Source: Risk Parity, Target: 60/40
rb = 3M-EDR, with trading costs UVT60/40
Total Source Return (gross of trading costs) 5.75
Leverage 2.66Excess Borrowing Return 1.49
Levered Excess Borrowing Return 3.97
Magnified Source Return 9.72
Volatility of Leverage 7.7212Volatility of Excess Borrowing Return 4.2219Correlation(Leverage,Excess Borrowing Return) -0.0566
Covariance(Leverage,Excess Borrowing Return) -1.84
Source Trading Costs -0.07
Leverage-Induced Trading Costs -0.96
Total Levered Return (arithmetic) 6.85
Compounded Arithmetic Return (gross) 1.0707Variance Correction 0.9934
Variance Drag -0.48
Approximation Error 0.00
Total Levered Return (geometric) 6.37
Table 2: Performance attribution of the realized geometric return of the levered strategy
UVT60/40 in terms of its source portfolio, risk parity, over the period January 1929–
December 2012. The performance attribution was based on Formulas (6) and (8). Bor-
rowing was at the Eurodollar deposit rate and trading costs were based on the linear
model in Appendix C. Arithmetic returns were estimated from monthly data and annu-
alized by multiplication by 12; they are displayed in percent. Geometric returns are also
displayed in percent and were annualized by (1 +G[r])12 − 1. Formulas corresponding to
the words in the performance attribution are presented in Table 9.
4 Assumptions about Transaction Costs and Their
Impact on Empirical Results
The return calculations in our empirical examples relied on assumptions about transac-
tion costs over our study period, 1929–2012. Comparisons between levered and unlevered
strategies were sensitive to these assumptions, but comparisons between strategies that
13
were comparably levered were much less sensitive to them. For transparency, we include
the details of our assumptions about transaction costs in Appendices B and C. Here, we
explain some of the reasoning that led to the choices we made, and we discuss the impact
of our choices on the results.
One guideline is that trading became less expensive over time during the study period,
so we assessed a greater cost to turnover at the beginning of the period than the end.
Specifically, we assumed that the portfolio was rebalanced monthly18 and that trading
cost 1% of the dollar amount of a trade between 1929 and 1955, .5% between 1956 and
1971, and .1% between 1972 and 2012. Since turnover tended to be higher in a levered
strategy than in an unlevered strategy, higher trading costs tended to do more damage
to a levered strategy than to an unlevered strategy.
As a borrowing rate, we used the 3-month Eurodollar deposit rate, for which we
had data back to the beginning of 1971. Prior to 1971, we used the 3-month T-bill rate
plus a spread of 60 basis points, which was 40 basis points less than the average spread
between the Eurodollar deposit rate and the T-bill rate between 1971 and 2012. This
choice improved the performance of our levered strategies relative to what they would
have been had we used the average spread. Of course, a lower borrowing rate would have
further improved the performance of the levered strategies.19 Since the levered strategies
involved borrowing and the unlevered strategies did not, there were more assumptions
underlying the empirical results for levered strategies than for unlevered strategies. As
a consequence, our uncertainty about results for levered strategies was greater than for
unlevered strategies.
It would, of course, have been possible to include empirical results based on a wider
range of assumptions about transaction costs. However, that would have been misleading
since it would have conveyed the impression that we had done a thorough study of the
issue. We did not. We chose a streamlined approach of providing examples based on
single set of assumptions that are consistent with published literature and that rely on
readily available data. The purpose of these examples is to illustrate the efficacy of our
performance attribution framework. We encourage practitioners and scholars to apply our
framework using their own estimates of trading and borrowing costs in order to evaluate
strategies and to facilitate the decision to lever.
18In practice, trading costs can be reduced by reducing the frequency or completeness of rebalancing,
at the cost of introducing tracking error. Further, trading costs may be higher for some asset classes than
for others. However, in our empirical examples, financing costs were more important than trading costs.19We considered using 1-month rates, but that would have engendered a more complex extrapolation
since the 1-month T-bill rate began only in 2001. Note that the difference between the 1-month and
3-month Eurodollar deposit rates averaged 20 basis points between 1971 and 2013. This was offset by the
40 basis points we subtracted in our extrapolation.
14
5 Benchmarks for a Levered Strategy
5.1 Fully Invested Benchmarks
Table 3 reports annualized arithmetic and geometric return, volatility and Sharpe ratio to
UVT60/40, its source, and its target. Because UVT60/40 was levered, while the source and
target were not, these comparisons were subject to uncertainty about historic financing
and trading costs. UVT60/40 had annualized geometric return only 63 basis points higher
than the source portfolio, unlevered risk parity.20 At the same time, the source portfolio
had a much lower volatility (4.20%). As a result, UVT60/40 had a Sharpe ratio of 0.29,
compared to 0.52 for unlevered risk parity. Note that the high Sharpe ratio of unlevered
risk parity was obtained at the cost of low expected return.
60/40 and UVT60/40 had essentially equal volatilities. Under our assumptions on his-
toric financing and trading costs, 60/40 delivered an annualized geometric return of 7.77%
and a realized Sharpe ratio of 0.40, while the analogous figures for UVT60/40 were 6.37%
and 0.29. Investors who are considering an investment in risk parity or any levered strat-
egy can populate Tables 2 and 3 with their forward-looking estimates of the components
of strategy return. This analysis can inform the decision to invest in a levered strategy
instead of the fully invested source or target portfolio.
5.2 Fixed Leverage and Conditional Leverage Benchmarks
In this section, we focus on comparisons of realized returns among levered strategies
that were constructed in different ways. These comparisons were less sensitive to the as-
sumptions on historical financing and trading costs. Like any volatility targeting strategy,
UVT60/40 was dynamically levered. However, as we saw in Section 3.2, the covariance
between leverage and excess borrowing return diminished annualized arithmetic return
by 1.84%. Deeper insight into this cost is provided in Table 2, which decomposes these
covariances into products of correlation and standard deviations. Note that the magni-
tude of the correlation between leverage and excess borrowing return was small: -0.056.
Figure 2 shows rolling 36-month estimates of the correlation between leverage and excess
borrowing return, and indicates that the sign of the correlation flipped repeatedly at short
horizons. At investment horizons of three to five years, the main effect of the covariance
term appeared to be to add noise to the returns.
20Note that the annualized geometric return of the source portfolio, 5.74%, slightly exceeded 5.68%,
the annualized arithmetic return of the source portfolio, net of trading costs. This is an artifact of
the annualization procedures for arithmetic and geometric return. The source portfolio had monthly
arithmetic return of 47.3 basis points, net of transaction costs. The latter was annualized by multiplying
by 12: 12×0.473% = 5.68%. Annualized geometric return takes into account compounding: 1.0047312−1 =
5.83%. The variance drag reduced this by 9 basis points to 5.74%. The variance drag on the source return
was much smaller than the variance drag on the levered portfolios, because the source portfolio was so
much less volatile and the variance drag is quadratic in volatility.
15
Table 3: Historical PerformanceSample Period: 1929-2012 Arithmetic Geometric Average Volatility Arithmetic Sharpe Skewness ExcessSource: Risk Parity, Target: 60/40 Total Total Leverage Excess Ratio Kurtosis
rb = 3M-EDR Return Return Return
60/40 8.18 7.77 1.00 11.58 4.69 0.40 0.19 7.44Risk Parity 5.68 5.74 1.00 4.20 2.20 0.52 0.05 4.92
UVT60/40 6.85 6.37 3.66 11.54 3.37 0.29 -0.43 2.23
Table 3: Annualized arithmetic and geometric returns, volatility and Sharpe ratio, of
UVT60/40 (risk parity levered to an unconditional volatility target of 11.59%, the realized
volatility of 60/40), the source portfolio (unlevered risk parity), and the volatility target
(60/40) over the period 1929–2012. Arithmetic returns were estimated from monthly data
and annualized by multiplication by 12; they are displayed in percent. Geometric returns
are also displayed in percent and were annualized by (1 + G[r])12 − 1. Volatility was
measured from monthly returns and annualized by multiplying by√
12. Sharpe ratios
were calculated using annualized excess return and annualized volatility.
−1
−0.8
−0.6
−0.4
−0.2
0
0.2
0.4
0.6
0.8
1
Cor
rela
tion
Source: Risk Parity, Target: 60/40 (1929−2012)Correlation Between Leverage and Excess Borrowing Return
1940
0930
1949
0630
1958
0331
1966
1230
1975
0930
1984
0629
1993
0331
2001
1231
2010
0930
UVT60/40
−120
−100
−80
−60
−40
−20
0
20
40
60
Cov
aria
nce
Source: Risk Parity, Target: 60/40 (1929−2012)Covariance Between Leverage and Excess Borrowing Return
1940
0930
1949
0630
1958
0331
1966
1230
1975
0930
1984
0629
1993
0331
2001
1231
2010
0930
UVT60/40
Figure 2: Correlation of excess borrowing return and leverage for UVT60/40, risk parity
levered to match the realized volatility of 60/40 fixed mix over the period 1929–2012.
Correlation was computed from monthly data using a trailing 36-month window.
When leverage is fixed, the covariance between leverage and excess borrowing return
must be zero. We consider two fixed leverage strategies: FLT60/40,λ matched the average
leverage of UVT60/40, but had higher volatility, while FLT60/40,σ matched the volatility of
UVT60/40 but had lower leverage.
Another alternative to UVT is a conditional volatility targeting strategy. CVT60/40
16
levered fully invested risk parity so that the projected volatility (based on the previous 36
months returns) equalled the volatility of the target 60/40 over the previous 36 months.21
Table 4 provides performance attributions for UVT60/40, FLT60/40,λ, FLT60/40,σ and
CVT60/40. Note that each column of Table 4 is a version of Table 2 applied to one of our
four levered strategies. All four levered strategies made use of the same source portfolio,
and hence had the same source arithmetic return. Leverage contributed substantially and
at roughly the same level to the magnified source return of UVT60/40, FLT60/40,λ and
CVT60/40, since those three strategies had similar average leverage. The contribution to
the return of FLT60/40,σ was significantly lower because that strategy had lower average
leverage. The covariance term reduced the annualized arithmetic return of UVT60/40 by
1.84%, but led to a much smaller reduction in the return of CVT60/40 and, by design, had
no effect on the return of the two FLT strategies. Trading costs reduced the return of
UVT60/40 and CVT60/40 by about 95 basis points, but had a smaller effect on the two FLT
strategies.22 The variance drag reduced the geometric returns of UVT60/40, FLT60/40,σ and
CVT60/40 by similar amounts, since these strategies had similar variances; the effect on
FLT60/40,λ was greater as a result of its higher volatility. When all the effects were taken
into account, the geometric returns of FLT60/40,λ, FLT60/40,σ and CVT60/40 exceeded the
geometric return of UVT60/40 by 192, 125 and 66 basis points, respectively.
21CVT60/40 was introduced in Anderson et al. (2012).22As discussed in Section 5.3 below, even maintaining a fixed leverage requires trading. It is possible
in principle that the trading needed to adjust leverage to meet a volatility target could offset some of the
trading required to maintain fixed leverage, but this strikes us as unlikely in typical situations. Had we
assumed lower trading costs, it would have narrowed the gap in trading costs among the strategies, but
not changed the ranking of those costs.
17
Table 4: Performance AttributionSample Period: 1929-2012Source: Risk Parity, Target: 60/40
rb = 3M-EDR, with trading costs UVT60/40 FLT60/40,λ FLT60/40,σ CVT60/40
Total Source Return (gross of trading costs) 5.75 5.75 5.75 5.75
Leverage 2.66 2.69 1.75 2.31Excess Borrowing Return 1.49 1.49 1.49 1.49
Levered Excess Borrowing Return 3.97 4.02 2.61 3.45
Magnified Source Return 9.72 9.77 8.37 9.20
Volatility of Leverage 7.7212 0.0000 0.0000 5.0791Volatility of Excess Borrowing Return 4.2219 4.2219 4.2219 4.2219Correlation(Leverage,Excess Borrowing Return) -0.0566 0.0000 0.0000 -0.0299
Covariance(Leverage,Excess Borrowing Return) -1.84 0.00 0.00 -0.64
Source Trading Costs -0.07 -0.07 -0.07 -0.07
Leverage-Induced Trading Costs -0.96 -0.51 -0.27 -0.93
Total Levered Return (arithmetic) 6.85 9.19 8.03 7.56
Compounded Arithmetic Return (gross) 1.0707 1.0959 1.0833 1.0783Variance Correction 0.9934 0.9881 0.9934 0.9926
Variance Drag -0.48 -0.91 -0.41 -0.53
Approximation Error 0.00 0.01 0.01 0.00
Total Levered Return (geometric) 6.37 8.29 7.62 7.03
Table 4: Performance attribution of the realized geometric return of the levered strategies
UVT60/40, FLT60/40,λ, FLT60/40,σ, and CVT60/40 in terms of their common source portfolio,
risk parity, over the period January 1929–December 2012. FLT60/40,λ had constant leverage
3.69, matching the average leverage of UVT60/40, while FLT60/40,σ had constant leverage
2.75, chosen to match the volatility of UVT60/40. The performance attribution was based
on Formulas (6) and (8). Borrowing was at the Eurodollar deposit rate and trading costs
were based on the linear model in Appendix C. Arithmetic returns were estimated from
monthly data and annualized by multiplication by 12; they are displayed in percent.
Geometric returns are also displayed in percent and were annualized by (1 +G[r])12 − 1.
Formulas corresponding to the words in the performance attribution are presented in
Table 9.
5.3 Attributes of Levered Strategies
The parameters of the UVT and two FLT levered strategies were set with foresight. The
dynamically levered strategy UVT60/40 was based on the realized volatility of a 60/40
fixed mix between January 1929 and December 2012. That volatility was known only at
period end even though it was used to make leverage decisions throughout the period. The
FLT60/40,λ leverage was set to match the average leverage of UVT60/40 and the FLT60/40,σ
leverage was set so that the volatility matched the volatility of UVT60/40.
CVT60/40, introduced in Section 5.2, did not rely on future information to set lever-
age.23 As a result, its realized volatility failed to match the realized volatility of the target.
At each monthly rebalancing, CVT60/40 was levered to match the volatility of the 60/40
fixed mix; both volatilities were estimated using a 36-month rolling window.
23The foresight in the definitions of UVT and the two FLT strategies allowed them to exactly match
their volatility or leverage targets, gross of trading costs. Since CVT60/40 did not rely on foresight, it
could not exactly match the realized target volatility, gross of trading costs. Both UVT and CVT60/40
volatility and FLT leverage were further affected by trading costs.
18
All else equal, UVT60/40, FLT60/40,λ, FLT60/40,σ and CVT60/40 called for additional
investment in the source portfolio when its price rose. A decline in the value of the source
portfolio reduced the net value of the levered portfolio, while keeping the amount borrowed
constant; leverage had increased, and rebalancing required selling the source portfolio to
return to leverage λ. Similarly, an increase in the value of the source portfolio resulted in
taking on more debt and using the proceeds to buy more of the source portfolio. In this
sense, the UVT, FLT and CVT strategies with λ > 1 were momentum strategies. UVT,
FLT and CVT strategies responded differently to changes in asset volatility; see Table 5.
Table 5: Strategy Responses to Changes in Market Conditions
Response:
Trigger FLT UVT CVT
Increase in Target Volatility no change no change ↑ leverageIncrease in Source Volatility no change ↓ leverage ↓ leverageIncrease in Price of Source buy source buy source buy source
Table 5: Responses of levered strategies to changes in market conditions.
5.4 Changing the Volatility Target
In this section, we explore the relationship between UVT and CVT strategies, and in par-
ticular their sensitivity to the volatility target. In addition to 60/40, we used the Market
Portfolio (i.e. the value-weighted portfolio of stocks and bonds, which has a higher volatil-
ity than 60/40) as the volatility target. UVTMKT and CVTMKT denote unconditionally
levered and conditionally levered risk parity strategies with the market as the volatility
target. Return comparisons of UVTMKT to CVTMKT and of UVT60/40 to CVT60/40 were
not sensitive to our assumptions on historical financing and trading costs, while the com-
parisons of UVTMKT to UVT60/40 and of CVTMKT to CVT60/40 were only slightly sensitive
to those assumptions.
Each term in the return attribution of the UVT risk parity strategies was sensitive
to the choice of MKT or 60/40 as the volatility target. By contrast, the magnified source
returns, covariance terms and trading costs of CVTMKT were quite similar to those of
CVT60/40; the only large difference between the two CVT strategies lay in the variance
drag. This finding indicates that CVT strategies were more stable than UVT strategies.
The geometric returns of UVTMKT (6.53%) and CVTMKT (6.52%) were virtually tied,
while CVT60/40 outperformed UVT60/40 by 63 basis points.24
24These findings do not support the assertion by Asness et al. (2013) that CVT is an inherently inferior
19
Table 6: Performance AttributionSample Period: 1929-2012Source: Risk Parity, Targets: VW Market, 60/40
rb = 3M-EDR, with trading costs UVTMKT UVT60/40 CVTMKT CVT60/40
Total Source Return (gross of trading costs) 5.75 5.75 5.75 5.75
Leverage 3.71 2.66 2.58 2.31Excess Borrowing Return 1.49 1.49 1.49 1.49
Levered Excess Borrowing Return 5.55 3.97 3.85 3.45
Magnified Source Return 11.30 9.72 9.60 9.20
Volatility of Leverage 9.9463 7.7212 5.3164 5.0791Volatility of Excess Borrowing Return 4.2219 4.2219 4.2219 4.2219Correlation(Leverage,Excess Borrowing Return) -0.0566 -0.0566 -0.0321 -0.0299
Covariance(Leverage,Excess Borrowing Return) -2.37 -1.84 -0.72 -0.64
Source Trading Costs -0.07 -0.07 -0.07 -0.07
Leverage-Induced Trading Costs -1.40 -0.96 -1.13 -0.93
Total Levered Return (arithmetic) 7.45 6.85 7.68 7.56
Compounded Arithmetic Return (gross) 1.0771 1.0707 1.0796 1.0783Variance Correction 0.9891 0.9934 0.9872 0.9926
Variance Drag -0.92 -0.48 -1.11 -0.53
Approximation Error 0.00 0.00 -0.05 0.00
Total Levered Return (geometric) 6.53 6.37 6.52 7.03
Table 6: Performance attribution of the realized geometric return of the levered strategies
UVTMKT, UVT60/40, CVTMKT and CVT60/40 in terms of their common source portfolio,
risk parity, over the period January 1929–December 2012. The performance attribution
was based on Formulas (6) and (8). Borrowing was at the Eurodollar deposit rate and
trading costs were based on the linear model in Appendix C. Arithmetic returns were
estimated from monthly data and annualized by multiplication by 12; they are displayed
in percent. Geometric returns are also displayed in percent and were annualized by (1 +
G[r])12 − 1. Formulas corresponding to the words in the performance attribution are
presented in Table 9.
5.5 Changing the Source Portfolio
Thus far, we have illustrated our performance attribution on a variety of risk parity
strategies that share a common source portfolio, unlevered risk parity. That allowed us to
isolate the impact of different leverage rules on performance.
In this section, we examine the impact of the source portfolio on performance: we
consider strategies that levered an index of US government bonds to target the volatility of
US equities. As in the previous examples, we consider both a dynamically levered volatility
targeting strategy, UVTBSTOCKS, as well as fixed leverage benchmarks, FLTBSTOCKS,λ
(with the same average leverage as UVTBSTOCKS) and FLTBSTOCKS,σ (with the same
volatility as UVTBSTOCKS). The details, presented in Table 7, were qualitatively similar
to what we saw for the risk parity strategies in Tables 4 and 6: an attractive magnified
source return was diminished substantially by transaction costs for all levered strategies
and by the covariance term for the dynamically levered strategy, UVTBSTOCKS. However,
since the source portfolio had lower volatility than unlevered risk parity, and the target
volatility was higher than that of 60/40 and the value-weighted market, leverage was
implementation of risk parity, compared to UVT.
20
higher and the effects were more dramatic.
The covariance term for UVTBSTOCKS was -4.23% per year, which imposed a larger
drag on return than did the covariance terms (-1.84% and -2.73%) for UVT60,40 and
UVTMKT. Despite the fact that the volatility target in UVTBSTOCKS was fixed, the leverage
was highly variable due to changes in the inverse of the volatility of the source portfolio
of U.S. Treasury bonds.25
The correlation between leverage and excess borrowing return to the source portfolio
was -.07. So as in the case of the dynamically levered risk parity strategies, a small
correlation resulted in a large return drag. The geometric returns to FLTBSTOCKS,λ and
FLTBSTOCKS,σ over our 84-year horizon were 5.93% and 6.94% per year. The geometric
return to UVTBSTOCKS over the same period was 1.7% per year.
25See Section 6 for an analysis of the covariance term from the standpoint of volatility targeting. Had
we made the unrealistic assumptions that financing was at the risk-free rate, and that trading costs were
zero, the two FLT strategies would still have easily outperformed the UVT strategy.
21
Table 7: Performance AttributionSample Period: 1929-2012Source: Bonds, Target: Stocks
rb = 3M-EDR, with trading costs UVTBSTOCKS FLTBSTOCKS,λ FLTBSTOCKS,σ
Total Source Return (gross of trading costs) 5.08 5.08 5.08
Leverage 6.43 6.49 4.80Excess Borrowing Return 0.82 0.82 0.82
Levered Excess Borrowing Return 5.29 5.34 3.95
Magnified Source Return 10.37 10.42 9.03
Volatility of Leverage 17.4159 0.0000 0.0000Volatility of Excess Borrowing Return 3.2711 3.2711 3.2711Correlation(Leverage,Excess Borrowing Return) -0.0742 0.0000 0.0000
Covariance(Leverage,Excess Borrowing Return) -4.23 0.00 0.00
Source Trading Costs 0.00 0.00 0.00
Leverage-Induced Trading Costs -2.59 -1.62 -0.93
Total Levered Return (arithmetic) 3.55 8.80 8.10
Compounded Arithmetic Return (gross) 1.0361 1.0916 1.0841Variance Correction 0.9820 0.9707 0.9823
Variance Drag -1.80 -2.84 -1.61
Approximation Error -0.05 -0.02 0.00
Total Levered Return (geometric) 1.70 5.93 6.49
Table 7: Performance attribution of the realized geometric return of the levered strategies
UVTBSTOCKS, FLTBSTOCKS,λ and FLTBSTOCKS,σ in terms of their common source port-
folio, U.S. Treasury bonds, over the period January 1929–December 2012. UVTBSTOCKS
was levered to the volatility of stocks (18.93%) over the period 1929–December 2012.
FLTBSTOCKS,λ had fixed leverage 8.72, equal to the average leverage of UVTBSTOCKS;
FLTBSTOCKS,σ had fixed leverage and the same volatility 22.47% as UVTBSTOCKS. The
performance attribution was based on Formulas (6) and (8). Borrowing was at the Eu-
rodollar deposit rate and trading costs were based on the linear model in Appendix C.
Arithmetic returns were estimated from monthly data and annualized by multiplication
by 12; they are displayed in percent. Geometric returns are also displayed in percent
and were annualized by (1 + G[r])12 − 1. Formulas corresponding to the words in the
performance attribution are presented in Table 9.
5.6 Historical Performance of the Various Levered and Fully
Invested Strategies
Table 8 summarizes the historical performance of our source portfolios (unlevered risk
parity and U.S. Treasury Bonds), volatility targets (fully invested 60/40, value-weighted
market, and stocks) and the various levered strategies considered in this paper. Unlev-
ered risk parity has the highest Sharpe ratio (0.52), followed closely by U.S. Treasury
Bonds (0.49). However, both exhibited low volatility and low excess return, making them
unattractive as asset allocations for most investors.26 Levered strategies are attractive as
an asset allocation only if the Sharpe ratio survives leverage.
26Of course, bonds are often used as one asset class in an asset allocation, such as 60/40 or the value-
weighted market portfolio. 60/40 has been widely used as an asset allocation, and risk parity has been
proposed as an alternative asset allocation; see, for example, Asness et al. (2012).
22
As shown in Table 8, the Sharpe ratios of the levered strategies were all lower than
the Sharpe ratios of their source portfolios. This highlights a fact that is well-known but
often neglected: outside of an idealized setting, the Sharpe ratio is not leverage invariant.
In this article, we highlight two features of a levered strategy that contribute to the
difference between its Sharpe ratio and the Sharpe ratio of its source portfolio. The first is
transaction costs. Both leverage-induced trading costs and financing costs diminish Sharpe
ratio; see Anderson et al. (2012, Formula 1). The second is the covariance term. Since the
covariance term was negative in the examples considered in this article, it lowered the
Sharpe ratios of the dynamically levered strategies relative to the Sharpe ratios of their
source portfolios and comparably calibrated fixed levered strategies. However, as indicated
in Figure 2, the correlation between leverage and the return to the source portfolio, which
is the driver of the covariance term, can be highly unstable at horizons of three to five
years. So unless a leverage-seeking investor has a specific reason to believe this correlation
will be positive over a particular period for a particular dynamically levered strategy, or
unless he or she enjoys the coin-flip-like risk illustrated in Figure 2, that investor may
prefer a fixed leverage strategy.
Table 8: Historical PerformanceSample Period: 1929-2012 Arithmetic Geometric Average Volatility Arithmetic Sharpe Skewness ExcessSource: Various, Target: Various Total Total Leverage Excess Ratio Kurtosis
rb = 3M-EDR Return Return Return
60/40 8.18 7.77 1.00 11.58 4.69 0.40 0.19 7.44VW Market 8.12 7.24 1.00 14.93 4.63 0.31 0.61 14.39Stocks 10.43 9.00 1.00 18.93 6.95 0.37 0.18 7.46Risk Parity 5.68 5.74 1.00 4.20 2.20 0.52 0.05 4.92Bonds 5.08 5.14 1.00 3.26 1.59 0.49 0.03 4.74
UVT60/40 6.85 6.37 3.66 11.54 3.37 0.29 -0.43 2.23
FLT60/40,λ 9.19 8.29 3.69 15.53 5.70 0.37 -0.01 4.78
FLT60/40,σ 8.03 7.62 2.75 11.57 4.54 0.39 0.00 4.80
CVT60/40 7.56 7.03 3.31 12.22 4.07 0.33 -0.41 7.13
UVTMKT 7.45 6.53 4.71 14.88 3.97 0.27 -0.44 2.23CVTMKT 7.68 6.52 3.58 16.13 4.19 0.26 -0.75 15.62
UVTBSTOCKS 3.55 1.70 7.43 19.10 0.07 0.00 -0.55 4.75FLTBSTOCKS,λ 8.80 5.93 7.49 24.47 5.31 0.22 -0.08 4.68FLTBSTOCKS,σ 8.10 6.49 5.80 18.95 4.61 0.24 -0.07 4.66
Table 8: Annualized arithmetic and geometric returns, volatility and Sharpe ratio, of the
source portfolios (unlevered risk parity and U.S. Treasury bonds, volatility targets (fully
invested 60/40, value-weighted market, and stocks) and the various levered strategies
considered in this paper, over the period 1929—2012. Arithmetic returns were estimated
from monthly data and annualized by multiplication by 12; they are displayed in percent.
Geometric returns are also displayed in percent and were annualized by (1 +G[r])12 − 1.
Volatility was measured from monthly returns and annualized by multiplying by√
12.
Sharpe ratios were calculated using annualized excess arithmetic return and annualized
volatility.
23
6 The Covariance Term, Revisited
The most novel part of our analysis is its focus on the covariance between leverage and
excess borrowing return. In this section, we examine the covariance term from the stand-
point of volatility targeting. We have already noted that leverage reduces the Sharpe
ratio if the borrowing rate exceeds the risk-free rate, or if trading incurs costs. However,
in a multi-period setting, leverage has an impact on Sharpe ratio even in the absence of
those market frictions, via the covariance term. In order to focus on the covariance term,
we make the highly unrealistic assumptions that borrowing is at the risk-free rate (i.e.
rb = rf ), which is fixed, and that trading costs are zero. We find that applying UVT
leverage does change the Sharpe ratio, even under these assumptions.27 Variable leverage,
as used in UVT, is “an unintended market-timing strategy.”28
Under these unrealistic assumptions, the excess return of the levered strategy is given
by:
rL − rf = λ(rS − rf
)(9)
Suppose we pick a fixed volatility target V ; then we must set λ = V
volatility of source.
Thus, we have:
Sharpe ratio of levered strategy
=E[rL − rf
]V
=E[λ(rS − rf
)]V
=E[
V
volatility of source×(rS − rf
)]V
=E[V × rS−rf
volatility of source
]V
= E
[rS − rf
volatility of source
]= E
[rS − rf
]E
[1
volatility of source
]+ cov
(rS − rf , 1
volatility of source
)(10)
Formula (10) makes it clear that a covariance term will be present in the Sharpe ratio of
27This issue has been misunderstood in the published literature. For example, Asness et al. (2013)
wrote, “Scaling the returns to any stable risk target (or not scaling them at all) cannot mathematically
affect the Sharpe ratio, or the t-statistic of the alpha of our levered portfolios, because we are multiplying
the return stream by a fixed constant.” Their analysis conflated single-period models with multi-period
models, and misstated the construction of the UVTMKT strategy used in Asness et al. (2012).28Asness et al. (2013) asserted that variable volatility, rather than variable leverage, is “an unintended
market-timing strategy.”
24
any strategy that involves levering a source portfolio of variable volatility to a volatility
target. Our empirical examples show that the covariance in Formula (5) has a material
effect on realized return and realized Sharpe ratio. It follows that the covariance in Formula
(10) has a material effect on realized return and realized Sharpe ratio. Recall that the
leverage was especially volatile in the levered strategy UVTBSTOCKS, which levered a
source portfolio of U.S. Treasury bonds to the volatility of stocks; even though the target
volatility was constant, the leverage was very volatile precisely because the inverse of the
volatility of bonds was high.
7 Conclusion
In this article, we developed a platform that supports both backward-looking performance
attribution and forward-looking investment decisions concerning levered strategies. Specif-
ically, in Formula (6), we expressed the difference between arithmetic expected return to
a levered strategy portfolio and its source portfolio as a sum of four terms:
E[rL]
= E[rS]
+ E [λ− 1]E[rS − rb
]+ cov(λ, rS − rb)−
(E[rTCS
]+ E
[rTCL
]).
The first two terms, whose sum we have called magnified source return, are the ones
that most easily come to mind in the context of a levered strategy. However, as we
have shown empirically, other factors have a material effect on the cumulative return to
a levered strategy. These include the covariance of leverage with the excess borrowing
return, trading costs and compounding effects.
Formula (6) accounted for both the covariance term and transaction costs. However,
it neglected the effect of compounding, which imposes a variance drag on cumulative
return that is not captured in arithmetic expected return. If the levered strategy has
high volatility, the variance drag may be substantial. Hence a more accurate decision rule
depends on geometric expected return in Formula (8):
G[r] ∼ (1 + E[r])e−var(r)
2 − 1.
We used Formulas (6) and (8) to examine the realized performance of fixed leverage
(FLT) strategies and two dynamically levered strategies: unconditional volatility targeting
(UVT) and conditional volatility targeting (CVT). Some scholars have expressed the view
that CVT strategies are poor alternatives to UVT strategies;29 this view is not supported
by the results reported in Tables 4 and 6. In fact, it is the leverage that was implicitly
determined by the volatility targets in UVT60/40 and CVT60/40, and not the volatility itself,
that interacted with the return to the source portfolio to determine strategy performance.
In our 1929–2012 period, CVT60/40 outperformed UVT60/40. Future work is required to
determine whether the sign of the covariance term might be predictable at longer horizons.
29See Asness et al. (2013).
25
In the examples we considered, the cumulative effects of borrowing and trading costs,
the variance drag and the covariance term offset much of the benefit of return magnifica-
tion. Leverage, both fixed and dynamic, substantially lowered Sharpe ratios. In addition,
dynamic leverage added noise to returns. Over our 84-year time horizon, fixed leverage
strategies outperformed volatility-targeting strategies, and levered strategies had lower
Sharpe ratios than their unlevered source portfolios.
Asness et al. (2012) argued that risk parity (levered to the volatility of the market)
outperformed 60/40 over a long horizon;30 our analysis does not support this.31 Risk
parity performed relatively well over the period 2008–2012, which featured Fed-supported
interest rates that were extraordinarily low by historical standards. But that need not
indicate how risk parity will perform in other regimes. Rising interest rates tend to raise
the cost of funding a levered strategy and lower the prices of bonds in risk parity portfolios
at the same time. Rising interest rates also have the potential to limit corporate profits
and thereby exert downward pressure on equity prices. These considerations should be
incorporated into any decision to lever low-risk portfolios when interest rates are unusually
low.
Appendices
A Related Literature
A.1 CAPM
Finance continues to draw heavily on the Capital Asset Pricing Model (CAPM) developed
in Treynor (1962), Treynor and Black (1976), Sharpe (1964), Lintner (1965a), Lintner
(1965b), Mossin (1966), and extended in Black and Litterman (1992).32 In the CAPM,
leverage is a means to adjust the level of risk in an efficient portfolio and nothing more. In
contrast, Markowitz (2005) illustrated another facet of leverage in the context of a market
composed of three coconut farms. In this disarmingly simple example, some investors
were leverage-constrained and others were not. The market portfolio was mean-variance
inefficient; as a result, no mean-variance investor would choose to hold it, and expected
returns of assets did not depend linearly on market betas.
30It is reasonable to compare the performance of the value-weighted market to risk parity levered to the
volatility of the market, and to compare the performance of 60/40 to risk parity levered to the volatility
of 60/40. However, it does not seem reasonable to compare the performance of 60/40 to that of risk parity
levered to the volatility of the market; we are grateful to Patrice Boucher for this insight.31See, however, the discussion in Section 4.32A history of the CAPM elucidating Jack Treynor’s role in its development is in French (2003).
26
A.2 Measurement of Risk and Nonlinearities
An impediment to a clear understanding of leverage may be the way we measure its
risk. Standard risk measures such as volatility, value at risk, expected shortfall, and beta
scale linearly with leverage. But as we know from the collapse of Long Term Capital,
the relationship between risk and leverage can be non-linear; see, for example, Jorion
(2000). Follmer and Schied (2002) and Follmer and Schied (2011, Chapter 4) described
risk measures that penalize leverage in a super-linear way. Recent experience suggests
that these measures may be useful in assessing the risk of levered strategies.
One contribution of this paper is to explain how the interaction between leverage
and market frictions creates specific nonlinearities in the relationship between leverage
and return. Understanding these specific nonlinearities provides a practical framework to
guide the decision on whether and how to lever.
A.3 Motivations for Leverage
If investors are overconfident in their predictions of investment returns, they may find
leverage attractive because it magnifies the returns when times are good, and because
they underestimate the risk of bad outcomes.33
Perfectly rational investors may also be attracted to leverage by the low risk anomaly,
the apparent tendency of certain low-risk portfolios to have higher risk-adjusted return
than high-risk portfolios. An investor who believes in the low risk anomaly will be tempted
to lever low-risk portfolios, in the hope of achieving high expected returns at acceptable
levels of risk.
In a CAPM world, investors with below-average risk aversion will choose to lever the
market portfolio.34 The low risk anomaly provides a rational argument for investors with
typical risk aversion to use leverage. Indeed, the low risk anomaly is arguably the only
rational argument for an investor to use leverage in an investment portfolio composed
of publicly traded securities.35 Differences in risk aversion could explain some investors
choosing higher expected return at the price of higher volatility, but there is little reason
for a rational investor to choose leverage unless the source portfolio being levered offers
superior risk-adjusted returns, at a volatility below the investor’s risk tolerance.
33A positive relationship between overconfident CEOs and firm leverage is documented in Malmendier
et al. (2011). Shefrin and Statman (2011) identified excessive leverage taken by overconfident bankers as
a contributor to the global financial crisis.34Note, however, that the market portfolio in CAPM includes bonds and other risky asset classes,
rather than just stocks. Levered strategies include the use of margin, and futures and other derivatives,
to assemble levered equity-only portfolios, which behave quite differently from levered portfolios in CAPM.35There are, of course, other rational arguments for using leverage in other contexts. The leverage
provided by a mortgage may be the only feasible way for a household to buy a house, which provides a
stream of consumption benefits and tax advantages in addition to facilitating an investment in the real
estate market. Companies leverage their shareholder equity with borrowing to finance operations, for a
variety of reasons, including differences in risk aversion, informational asymmetries, and tax implications.
27
A.4 Levered Low-Risk Strategies
Low-risk investing refers to a diverse collection of investment strategies that emphasize low
beta, low idiosyncratic risk, low volatility or downside protection. The collection of low-
risk strategies includes broad asset allocations, but it also includes narrower strategies
restricted to a single asset class. An early reference to low-risk investing is Markowitz
(1952) who commented that a minimum-variance portfolio is mean-variance optimal if
all assets returns are uncorrelated and have equal expectations. But low-risk strategies
typically require leverage in order to meet expected return targets. In an exploration
of this idea, Frazzini and Pedersen (2013) echoed some of the conclusions in Markowitz
(2005), and they complemented theory with an empirical study of an implicitly levered
equity risk factor that was long low-beta stocks and short high-beta stocks. This factor
descended from Black et al. (1972), which provided evidence that the CAPM may not
properly reflect market behavior.
A.5 Empirical Evidence on Levered Low-Risk Investing
There is a growing empirical literature indicating that market frictions may prevent in-
vestors from harvesting the returns promised by a frictionless analysis of levered low-risk
strategies. Anderson et al. (2012) showed that financing and trading costs can negate the
abnormal profits earned by a levered risk parity strategy in a friction-free market. Li et al.
(2014) and Fu (2009) showed that market frictions may impede the ability to scale up the
return of low-risk strategies through leverage.36
Asset allocation that is based on capital weights has a long and distinguished his-
tory; see, for example Graham (1949) and Bogle (2007). However, rules-based strategies
that allocate risk instead of, or in addition to, capital are of a more recent vintage. Risk-
based investing is discussed in Lortscher (1990), Kessler and Schwarz (1996), Qian (2005),
Clarke et al. (2011), Shah (2011), Sefton et al. (2011), Clarke et al. (2013), Anderson et al.
(2012), Cowan and Wilderman (2011), Bailey and de Prado (2012), Goldberg and Mah-
moud (2013) and elsewhere. Strategies that target volatility are also gaining acceptance,
although the literature is still sparse. Goldsticker (2012) compared volatility targeting
strategies to standard allocations such as fixed-mix, and found that the relative perfor-
mance of the strategies was period dependent.
A.6 The Effect of Leverage on Markets
Another important question is the extent to which leverage may contribute to market
instability. See, for example, Brunnermeier and Pedersen (2009), Adrian and Shin (2010)
and Geanakoplos (2010). We do not address that question here, as we restrict our analysis
36Ross (2004) provided an example of the limits to arbitraging mispricings of interest-only strips of
mortgage backed securities.
28
to the effect of leverage on the return of investment strategies, taking the distribution of
the underlying asset returns as given.
A.7 Arithmetic versus Geometric Return
Despite the large literature on the importance of compounding to investment outcomes,
analyses of investment strategies are often based on arithmetic expected return. Back-
ground references on compounding and geometric return include Fernholz (2002) and
MacLean et al. (2011). Perold and Sharpe (1988) discussed how the interplay among
volatility, rebalancing and compound return causes a fully-invested fixed-mix or portfolio-
insurance strategy to behave differently from a buy-and-hold strategy with the same initial
mix. Booth and Fama (1992) worked out the relationship between the compound return
to a fixed-mix portfolio and its constituents, and their results were applied to portfolios
that include commodities in Willenbrock (2011). Markowitz (2012) compared six different
mean-variance approximations to geometric return.
B Data
The results presented in this paper were based on CRSP stock and bond data from
January of 1929 through December of 2012. The aggregate stock return is the CRSP
value weighted market return (including dividends) from the table Monthly Stock–Market
Indices (NYSE/AMEX/NASDAQ) – variable name vwretd. The aggregate bond return
was the face value outstanding (cross-sectionally) weighted average of the unadjusted
return for each bond in the CRSP Monthly Treasury (Master) table. In this table, the
variable name for the unadjusted return is retnua and for the face value outstanding is
iout1r. All bonds in the table were used, provided the values for both retnua and iout1r
were not missing.
The proxy for the risk-free rate was the USA Government 90-day T-Bills Secondary
Market rate, provided by Global Financial Data (http://www.globalfinancialdata.
com), covering the period from January of 1929 through December of 2012. The proxy for
the cost of financing leverage was the U.S. 3-Month Euro-Dollar Deposit rate, downloaded
from the Federal Reserve (http://www.federalreserve.gov/releases/h15/data.htm).
The 3-Month Euro-Dollar Deposit data is available from January of 1971 through Decem-
ber of 2012. Prior to January of 1971, a constant of 60 basis points was added to the
90-day T-Bill rate.37 Trading costs were calculated using the procedure described in Ap-
pendix C. We assumed the cost of trading was 100 basis points from 1926 to 1955, 50
basis points from 1956 to 1970, and 10 basis points from 1971 onward.
The construction of the unlevered and levered risk parity strategies was exactly as
37The average difference between the 90-day T-Bill Rate and the 3-Month Euro-Dollar Deposit Rate
from 1971 through 2012 was 102 basis points. So our estimate of 60 basis points was relatively conservative.
29
detailed in Anderson et al. (2012). The construction of the bonds levered to stock strategies
was the analogue for the case of a single asset class.
Anderson et al. (2012), following Asness et al. (2012), used the volatility of the market
as the target for risk parity. Here, we used the volatility of 60/40 as the target, because
it provides a more appropriate comparison to traditional strategies used by institutional
investors. The return of UVT strategies is particularly sensitive to the volatility target.
C Trading Costs
We estimate the drag on return that stems from the turnover-induced trading required
to maintain leverage targets in a strategy that levers a source portfolio S.
At time t, the strategy calls for an investment with a leverage ratio of λt. We make
the harmless assumption that the value of the levered strategy at t, denoted Lt, is $1.38
Then the holdings in the source portfolio, or assets, are At = λt. The debt at time t is
given by Dt = λt − 1.
We need to find holdings At+1 in the portfolio at time t+ 1 that are consistent with
the leverage target λt+1. This turns out to be a fixed point problem since the trading
costs must come out of the investor’s equity. Between times t and t + 1, the value of the
source portfolio changes from St to St+1 and the strategy calls for rebalancing to achieve
leverage λt+1. Just prior to rebalancing, the value of the investment is
A′t = λt(1 + rSt ), (11)
the liability has grown to D′t = (λt − 1)(1 + rbt ) and the investor’s equity is:
L′t = A′t −D′t= λt(1 + rSt )− (λt − 1)(1 + rbt ). (12)
Note that in Formulas (11) and (12), we use the source return rSt gross of trading costs
in the source portfolio.
Let wt = (wt1, . . . , wtn)> denote the vector of relative weights assigned to the n asset
classes in the source portfolio at time t, so that∑n
i=1wti = 1 for all t. Just prior to
rebalancing, the weights have changed to w′t = (w′t1, . . . , w′tn)>, where w′ti =
wti(1+rti)
1+rSt. At
time t + 1, the strategy is rebalanced according to its rules, which produces holdings of
At+1wt+1 in the n asset classes. We let xt = (xt1, . . . , xtn)> denote the vector of dollar
amounts of the changes in value due to rebalancing, so that:
xt = At+1wt+1 − A′tw′t. (13)
38This assumption is harmless in a linear model of trading costs, which we develop here. It would be
inappropriate for a realistic model of market impact.
30
If we assume a linear model, the cost of trading xt is κ‖xt‖1 =∑n
i=1 |xti| for some κ ≥ 0.
The cost reduces the investor’s equity to:
Lt+1 = L′t − κ‖xt‖1= λt(1 + rSt )− (λt − 1)(1 + rbt )− κ‖xt‖1. (14)
Now let
g(α) =α
Lt+1
− λt+1
=α
L′t − h(α)− λt+1,
where g(α) denotes the leverage implied by holding αwt+1 in the n assets, taking into
account the effect of trading costs on equity Lt+1, minus the desired leverage. Assuming
that g is defined on the whole interval [0, λt+1L′t], it is continuous, g(0) = −λt+1 < 0, and
g(λt+1Lt′), so by the Intermediate Value Theorem, there exists αt+1 such that g(αt+1) =
0.39 The value of αt+1 can readily be found by a bisection algorithm, which worked well
in all of the empirical situations studied in this paper.40
We set At+1 = αt+1, so the holdings of the n assets are given by At+1wt+1 = αt+1wt+1.
The reduction in return due to trading costs is given by:
rTC = κ‖αt+1wt+1 − A′tw′t‖1. (15)
We compute the trading cost incurred by the source portfolio, E[rTCS] in the same
way and define the trading cost due to leverage by
E[rTCL] = E[rTC]− E[rTCS]. (16)
D Geometric Return
In order to analyze the effects of compounding, Booth and Fama (1992) expressed con-
tinuously compounded return in terms of arithmetic return. We have chosen to analyze
the effects of compounding using the geometric average of monthly returns. Our Formula
(19) for the geometric average of monthly returns is somewhat simpler than the formula
for continuously compounded return in Booth and Fama (1992). Both derivations rely on
the second-order Taylor expansion approximation of the logarithm.
Let Lt denote the equity in a strategy at month t, where t = 0, 1, . . . , T .
39Typically, αt+1 is uniquely determined; if not, choose the largest value satisfying the equation.40If there is no αt+1 such that g(α) = 0, it means the equity of the strategy is so low that the transaction
costs in getting to the desired leverage wipe out the equity. We do not observe such severe drawdown in
our empirical examples, but clearly it would be possible with extreme leverage or a very volatile source
portfolio.
31
The correct ranking of realized strategy performance, taking compounding into ac-
count, is given by G[r], the geometric average of the monthly returns, minus one:
G [r] =
(LTL0
)1/T
− 1
=
[T−1∏t=0
Lt+1
Lt
]1/T− 1
=
[T−1∏t=0
(1 + rt)
]1/T− 1 (17)
Because the logarithm is strictly increasing, log (1 +G [r]) induces exactly the same
ranking of realized strategy returns as G[r]. It is a different ranking than the one induced
by E[r] and log (1 + E [r]), requiring a correction term involving var(r):
log (1 +G [r]) =1
T
T−1∑t=0
log (1 + rt)
∼ 1
T
T−1∑t=0
(rt −
(rt)2
2
)(18)
=1
T
T−1∑t=0
rt −1
T
T−1∑t=0
(rt)2
2
= E [r]− var (r) + (E (r))2
2
∼ log (1 + E [r])− var (r)
2(19)
G [r] ∼ (1 + E [r]) e−var(r)
2 − 1 (20)
Formulas (18) and (19) approximate the logarithm by its quadratic Taylor polynomial.
When rt > 0, the Taylor series for logarithm is alternating and decreasing in absolute
value for |rt| < 1, so the error in the approximation of log (1 + rt) in Formula (18) is
negative and bounded above in magnitude by |rt|3 /3 for each month t. When rt < 0, the
error is positive and may be somewhat larger than |rt|3 /3. Since the monthly returns are
both positive and negative, the errors in months with negative returns will substantially
offset the errors in months with positive returns, so the errors will tend not to accumulate
over time. The approximation error in annual geometric return was at most one basis
point in our risk parity examples (see Table 4) and five basis points in our levered bond
examples (see Table 7).
32
E Words and Formulas
Table 9 presents the formulas accompanying the words in our Performance Attribution
Tables 2, 4, 6 and 7.
Table 9: Performance AttributionDescription: Formula:
Total Source Return (gross of trading costs) E[rS] (gross of trading costs)
Leverage E[λ − 1]
Excess Borrowing Return E[rS − rb]
Levered Excess Borrowing Return E[λ − 1] · E[rS − rb]
Magnified Source Return E[rS] + E[λ − 1] · E[rS − rb]
Volatility of Leverage σ(λ)
Volatility of Excess Borrowing Return σ(rS − rb)
Correlation(Leverage,Excess Borrowing Return) ρ(λ, rS − rb)
Covariance(Leverage,Excess Borrowing Return) Cov(λ, rS − rb)
Source Trading Costs −E[rT CS ]
Leverage-Induced Trading Costs −E[rT CL]
Total Levered Return (arithmetic) E[rL]
Compounded Arithmetic Return (gross) (1 + E[rL]/1200)12
Variance Correction exp(−σ2rL /2)
Variance Drag [(1 + E[rL]/1200)12 · exp(−σ2rL /2) − 1] · 100 − E[rL]
Approximation Error G[rL] − [(1 + E[rL]/1200)12 · exp(−σ2rL /2) − 1] · 100
Total Levered Return (geometric) G[rL]
Table 9: Formulas corresponding to the words used in Performance Attribution Tables 2,
4, 6 and 7.
References
Tobias Adrian and Hyun Song Shin. Liquidity and Leverage. J. Finan. Intermediation,
19:418–437, 2010.
Robert M. Anderson, Stephen W. Bianchi and Lisa R. Goldberg. Will My Risk Parity
Strategy Outperform? Financial Analysts Journal, 68(6):75–93, 2012.
Clifford Asness, Andrea Frazzini and Lasse Heje Pedersen. Leverage Aversion and Risk
Parity. Financial Analysts Journal, 68(1):47–59, 2012.
Clifford Asness, Andrea Frazzini and Lasse Heje Pedersen. Will My Risk Parity Strategy
Outperform?: A Comment. Financial Analysts Journal, 69(2):12–15, 2013.
David H. Bailey and Marcos Lopez de Prado. Balanced Baskets: A New Approach to
Trading and Hedging Risks. Journal of Investment Strategies, 1(4):21–62, 2012.
33
Fischer Black, Michael C. Jensen and Myron Scholes. The Capital Asset Pricing Model:
Some Empirical Tests. In Michael C. Jensen, editor, Studies in the Theory of Capital
Markets, pages 79–121. Praeger Publishers Inc., 1972.
Fisher Black and Robert Litterman. Global Portfolio Optimization. Financial Analysts
Journal, September:28–43, 1992.
John C. Bogle. The Little Book of Common Sense Investing: The Only Way to Guarantee
Your Fair Share of Stock Market Returns (Little Books. Big Profits). Wiley, 2007.
David G. Booth and Eugene F. Fama. Diversification Returns. Financial Analysts Journal,
48(3):26–32, 1992.
Markus K. Brunnermeier and Lasse Heje Pedersen. Market Liquidity and Funding Liq-
uidity. Review of Financial Studies, 22(6):2201–2238, 2009.
Roger Clarke, Harindra De Silva and Stephen Thorley. Minimum Variance Portfolio
Composition. Journal of Portfolio Management, 37(2):31–45, 2011.
Roger Clarke, Harindra De Silva and Stephen Thorley. Minimum Variance, Maximum
Diversification and Risk Parity: An Analytic Perspective. Journal of Portfolio Man-
agement, 39(3):39–53, 2013.
David Cowan and Sam Wilderman. Rethinking Risk: What the Beta Puzzle Tells us
About Investing, 2011. GMO White Paper.
E. Robert Fernholz. Stochastic Portfolio Theory. Springer, New York, 2002.
Hans Follmer and Alexander Schied. Convex Risk Measures and Trading Constraints.
Finance and Stochastics, 6:429–427, 2002.
Hans Follmer and Alexander Schied. Stochastic Finance: An Introduction in Discrete
Time. Walter de Gruyter, third edition, 2011.
Andrea Frazzini and Lasse Heje Pedersen. Betting Against Beta. forthcoming in Journal
of Financial Economics, 2013.
Craig French. The Treynor Capital Asset Pricing Model. Journal of Investment Manage-
ment, 1(2):60–72, 2003.
Fangjian Fu. Idiosyncratic Risk and the Cross-Section of Expected Stock Returns. Journal
of Financial Economics, 91:24–37, 2009.
John Geanakoplos. The Leverage Cycle, 2010. Cowles Foundation Discussion Paper No..
1715R.
34
Lisa R. Goldberg and Ola Mahmoud. Risk Without Return. Journal of Investment
Strategies, 2(2):111–120, 2013.
Ralph Goldsticker. Volatility Targeting May Miss the Mark, 2012. BNY Mellon Asset
Management.
Benjamin Graham. The Intelligent Investor. Collins, 1949.
Philippe Jorion. Risk Management Lessons From Long-Term Capital Management. Eu-
ropean Financial Management, 2000.
Christoph Kessler and Gunter Schwarz. Investment management for private investors at
SBC - The benchmark, 1996. Swiss Bank Corporation / The Global 2nd Quarter.
Rob Kozlowski. Kansas City Employees pension fund commits to BlackRock for Risk
Parity. Pensions & Investments, June 20, 2013.
Xi Li, Rodney Sullivan and Luis Garcia-Feijoo. The Limits to Arbitrage and the Low
Risk Anomaly. Financial Analysts Journal, 70(1):52–63, 2014.
John Lintner. Securities, Prices, Risk and Maximal Gains from Diversification. The
Journal of Finance, 20(4):13–37, 1965a.
John Lintner. The Valuation of Risk Assets and the Selection of Risky Investments in
Stock Portfolios and Capital Budgets. Review of Economics and Statistics, 47(1):587–
615, 1965b.
Rudolf Lortscher. Investment Policy at SBC - basic premises and model, 1990. Swiss
Bank Corporation / The Global 10/90.
Leonard C. MacLean, Edward O. Thorp and William T. Ziemba. The Kelly Capital
Growth Investment Criterion. World Scientific, 2011.
Ulrike Malmendier, Geoffrey Tate and Jon Yan. Overconfidence and Early-Life Expe-
riences: The Effect of Managerial Traits on Corporate Financial Policies. Journal of
Finance, 66(5):1687–1783, 2011.
Harry Markowitz. Portfolio Selection. Journal of Finance, VII(1):77–91, 1952.
Harry Markowitz. Mean-Variance Approximations to the Geometric Mean. Annals of
Financial Economics, 7(1), 2012.
Harry M. Markowitz. Market Efficiency: A Theoretical Distinction and So What? Finan-
cial Analysts Journal, September/October:17–30, 2005.
Jan Mossin. Equilibrium in a Capital Asset Market. Econometrica, 34(4):768–783, 1966.
35
Lorenzo Naranjo. Implied Interest Rates in a Market with Frictions, 2009. Working paper,
New York University.
Andre F. Perold and William F. Sharpe. Dynamic Strategies for Asset Allocation. Fi-
nancial Analysts Journal, 44(1):16–27, 1988.
Edward Qian. Risk Parity Portfolios: Efficient Portfolios Through True Diversification,
2005. PanAgora.
Stephen Ross. Neoclassical Finance. Princeton University Press, 2004.
James Sefton, David Jessop, Giuliano De Rossi, Claire Jones and Heran Zhang. Low-Risk
Investing, 2011. UBS Investment Research.
Ronnie R. Shah. Understanding Low Volatility Strategies: Minimum Variance, 2011.
Dimensional Fund Advisors.
William Sharpe. Capital Asset Prices: A Theory of Market Equilibrium Under Conditions
of Risk. Journal of Finance, 19(3):425–442, 1964.
Hersh Shefrin and Meir Statman. Behavioral Finance in the Financial Crisis: Market
Efficiency, Minsky, and Keynes, 2011. Santa Clara University.
Rodney Sullivan. Speculative Leverage: A False Cure for Pension Woes. Financial Ana-
lysts Journal, 66(3):6–8, 2010.
Jack Treynor. Toward a Theory of Market Value of Risky Assets, 1962. Presented to the
MIT Finance Faculty Seminar in 1962, finally published in 1999 in Asset Pricing and
Portfolio Performance, Robert J. Korajczyk (editor) London: RiskBooks, 15–22.
Jack L. Treynor and Fischer Black. Capital Market Theory and the Pricing of Financial
Securities. In Stewart C. Myers, editor, Modern Developments in Financial Manage-
ment, pages 310–327. Praeger Publishers, 1976.
Scott Willenbrock. Diversification Return, Portfolio Rebalancing, and the Commodity
Return Puzzle. Financial Analysts Journal, 67(4):42–49, 2011.
36